Tom Lane wrote:
> Madison Kelly <linux@alteeve.com> writes:
>> Alvaro Herrera wrote:
>>> Please send along
>>> select xmin, xmax, ctid, cmin, cmax, datname from pg_database;
>
>> template1=# select xmin, xmax, ctid, cmin, cmax, datname from pg_database;
>> xmin | xmax | ctid | cmin | cmax | datname
>> ------+------+--------+------+------+------------
>> 383 | 0 | (0,1) | 0 | 0 | template1
>> 384 | 0 | (0,2) | 0 | 0 | template0
>> 386 | 0 | (0,3) | 0 | 0 | postgres
>> 659 | 0 | (0,10) | 0 | 0 | deadswitch
>> 3497 | 3625 | (0,35) | 0 | 0 | nexxia
>> (5 rows)
>
> So the "nexxia" row did get updated at some point, and either that
> transaction failed to commit or we've got some glitch that made this
> row look like it didn't. Have you used any "ALTER DATABASE" commands
> against nexxia?
>
> regards, tom lane
>
Nope.
Beyond the occasional ALTER COLUMN (few and always completed), the only
thing I do directly in the shell are pretty standard queries while
working out my program. Even then, the database is dropped and recreated
fairly regularly with backup copies from the server.
Madi
PS - If I've run into a PgSQL bug, is there anything I can provide to help?